Execute Notebook

Execute a Jupyter notebook with full lifecycle management.

Notebook: $ARGUMENTS

Options

• --mode: full (default), selective, papermill
• --cells: Specific cells (e.g., 0,5,10)
• --timeout: Per-cell timeout in seconds (default 300)
• --stop-on-error: Halt on first error
• --batch: Enable BATCH_MODE for interactive notebooks
• --save: Save outputs to original notebook

Execution by Kernel Type

Python (kernel: python3)

Local scripts (preferred for batch):

# Full execution via notebook_tools.py
python scripts/notebook_tools/notebook_tools.py execute {path} --timeout 300

# Cell-by-cell with verbose output via notebook_helpers.py
python scripts/notebook_tools/notebook_helpers.py execute {path} --verbose

# Specific cell execution
python scripts/notebook_tools/notebook_helpers.py execute {path} --cell 5 --timeout 60

MCP (for interactive debugging):

• manage_kernel(action="start", kernel_name="python3") + execute_on_kernel

.NET (kernel: .net-csharp)

• MCP cell-by-cell ONLY - Papermill and local scripts do NOT work with .NET kernels
• CRITICAL: Set working directory first with SetCurrentDirectory()
• See mcp-jupyter skill for detailed .NET patterns

GenAI notebooks

• Check .env configuration first (use /validate-genai)
• Use --batch for notebooks with interactive widgets
• Use dedicated scripts: python scripts/genai-stack/validate_notebooks.py

Error Handling

Report Format

Report: notebook name, kernel, mode, total/executed/successful/error cells, timing, error details.